The Human Rights Effects of World Bank Structural Adjustment,

Transcription

1 International Studies Quarterly (2006) 50, The Human Rights Effects of World Bank Structural Adjustment, M. RODWAN ABOUHARB Louisiana State University DAVID L. CINGRANELLI Binghamton University Does the implementation of a World Bank structural adjustment agreement (SAA) increase or decrease government respect for human rights? Neoliberal theory suggests that SAAs improve economic performance, generating better human rights practices. Critics contend that the implementation of structural adjustment conditions causes hardships and higher levels of domestic conflict, increasing the likelihood that regimes will use repression. Bivariate probit models are used to account for World Bank loan selection criteria when estimating the human rights consequences of structural adjustment. Using a global, comparative analysis for the period, we examine the effects of structural adjustment on government respect for citizens rights to freedom from torture, political imprisonment, extra-judicial killing, and disappearances. The findings show that World Bank SAAs worsen government respect for physical integrity rights. World Bank and International Monetary Fund (IMF) structural adjustment conditions require loan recipient governments to rapidly liberalize their economies. According to previous research, these economic changes often cause at least shortterm hardships for the poorest people in less developed countries. The Bank and IMF justify the loan conditions as necessary stimuli for economic development. However, research has shown that implementation of structural adjustment conditions actually has a negative effect on economic growth (Przeworski and Vreeland 2000; Vreeland 2003). While there has been less research on the human rights effects of structural adjustment conditions, most studies agree that the imposition of structural adjustment agreements (SAAs) on less developed countries worsens government human rights practices (Pion-Berlin 1984; McLaren 1988; Franklin 1997; Camp Keith and Poe 2000). This study focuses on the effects of structural adjustment conditions on the extent to which governments protect their citizens from extra-judicial killing, torture, disappearances, and political imprisonment. Authors note: This research was supported by a grant (No. SES ) from the Political Science Division of the National Science Foundation. Any opinions, findings, and conclusions or recommendations expressed in this paper are those of the author(s) and do not necessarily reflect the views of the National Science Foundation. The empirical tests were conducted using STATA 9. A list of countries used in the study, replication data and files for Tables 3 5 and Column IV of Table 6 are available from hhttp:// r 2006 International Studies Association. Published by Blackwell Publishing, 350 Main Street, Malden, MA 02148, USA, and 9600 Garsington Road, Oxford OX4 2DQ, UK.

2 234 The Human Rights Effects of World Bank Structural Adjustment The results of this study suggest that existing theories of repression should be revised to take greater account of transnational causal forces. Previous studies examining variations in the human rights practices of governments have concentrated almost exclusively on state-level characteristics such as wealth, constitutional provisions, or level of democracy (e.g., Mitchell and McCormack 1988; Poe and Tate 1994; Davenport 1996; Poe, Tate, and Camp Keith 1999; Davenport and Armstrong 2004). The dominant theoretical framework underlying this research argues that, other things being equal, repression will increase as regimes are faced with a domestic threat in the form of civil war or when a country is involved in international war (Poe, Tate, and Camp Keith 1999; Poe 2002:293; see also Gurr 1986; Davenport 1995). Other international factors besides involvement in international war such as the degree of integration into the global economy, sensitivity to international norms, and involvement with international financial institutions have received much less attention. 1 Empirically, this study advances our understanding of the human rights consequences of structural adjustment by correcting for the effects of selection. It is possible that the worsened human rights practices observed and reported in previous studies might have resulted from the poor economic conditions that led to the imposition of the structural adjustment conditions rather than the implementation of the structural adjustment conditions themselves. In other words, the human rights practices of loan recipient governments might have gotten worse whether or not a structural adjustment agreement (SAA) had been received and implemented. In addition, as our results will show, some of the factors that increase the probability of entering into a SAA, such as having a large population and being relatively poor, are also associated with an increased probability of human rights violations. For these reasons one must disentangle the effects of selection before estimating the human rights impacts of structural adjustment loans. In order to control for the effects of selection, a two-stage analysis was undertaken. In the first stage of the analysis, the factors affecting World Bank decisions concerning which governments receive SAAs were identified. In the second stage the impacts of entering into and implementing SAAs on government respect for human rights were examined. The first-stage results demonstrate that the Bank does give SAAs to governments that are poor and experiencing economic trouble, but the Bank also employs a wide variety of non-economic loan selection criteria. The non-economic selection criteria examined in the first stage of the analysis build upon and extend selection models developed in previous research on the economic effects of structural adjustment. This research project is the first to demonstrate that the Bank prefers to give loans to governments that provide greater protection for worker rights and physical integrity rights of their citizens. Earlier research had shown that democracies were at a disadvantage when negotiating a SAA from the IMF (Przeworski and Vreeland 2000; Vreeland 2003), a finding consistent with expectations generated by Putnam s (1988) theory of two-level games. Our findings provide evidence that democracies also are at a disadvantage when negotiating with the World Bank. After controlling for selection effects and other explanations of respect for physical integrity rights, the findings of the second-stage analysis show that the net effect of World Bank SAAs is to worsen government respect for physical integrity rights. Torture, political imprisonment, extra-judicial killing, and disappearances were all more likely to occur when a structural adjustment loan had been received and implemented. Governments that entered into SAAs with the World Bank actually 1 Some scholars have focused on transnational forces affecting human rights practices. For example, increased integration into the international economy has been associated with both worse (Meyer 1996, 1998) and better (Milner 2000; Richards, Gelleny, and Sacko 2001) protection of physical integrity rights by governments. Other studies have discussed the impacts of international nongovernmental organizations (Welch 1995) and even the diffusion of international norms (Keck and Sikkink 1998; Landman 2005) on the human rights practices of governments.

3 M. RODWAN ABOUHARB AND DAVID L. CINGRANELLI 235 improved their protection of physical integrity rights in the year the loan was received. Governments then reduced the level of respect for the physical integrity rights of their citizens during the years when structural adjustment conditions were imposed. This combination of findings suggests that governments seeking loans from the World Bank initially improved their human rights practices, possibly to impress Bank officials. However, the austerity measures required by the implementation of structural adjustment conditions led to a subsequent worsening of human rights practices by governments in loan recipient countries. The theoretical argument is that there are both direct and indirect negative effects of the implementation of structural adjustment conditions on government respect for physical integrity rights. Structural adjustment conditions almost always cause hardships for the poorest people in a society, because they necessitate some combination of reductions in public employment, elimination of price subsidies for essential commodities or services, and cuts in expenditures for health, education and welfare programs. These hardships often cause increased levels of domestic conflict that present substantial challenges to government leaders. Some governments respond to these challenges by becoming less democratic as in the case of Peru under President Fujimori in the 1980s (Di John 2005). 2 The results presented here, like those of numerous other studies, have shown that increased domestic conflict and decreased democracy are associated with higher levels of repression (e.g., Poe, Tate, and Camp Keith 1999). The case of Venezuela provides an illustration of the role of structural adjustment in producing increased domestic conflict, a weakened democratic system and repression. As Di John (2005:114) writes: A few weeks after the announcement of [structural adjustment] reforms, Venezuela experienced the bloodiest urban riots since the urban guerrilla warfare of the 1960s. The riots, known as the Caracazo, occurred in late February A doubling of gasoline prices, which were passed on by private bus companies, induced the outburst.... The riots that ensued were contained by a relatively undisciplined military response that left more than 350 dead in two days. Although Venezuela s democratic system has been maintained, over the period of this study, dissatisfaction with economic policies has played a part in three attempted coups, multiple general strikes, two presidential assassination attempts, and has led to several states of emergency being imposed. Even today, debate over structural adjustment policies in Venezuela remains heated. President Hugo Chavez sustains his popularity largely based on his opposition to the kind of unregulated economic liberalization advocated by the IMF and the Bank (Banks, Muller, and Overstreet 2003). The findings presented here have important policy implications. There is mounting evidence that national economies grow fastest when basic human rights are respected (Sen 1999; Kaufmann 2004; Kaufmann, Kraay, and Mastruzzi 2005). SAAs place too much emphasis on instituting a freer market and too little emphasis on allowing the other human freedoms necessary for rapid economic growth to take root and grow. By undermining the human rights conditions necessary for economic development, the Bank is damaging its own mission. Background While each structural adjustment program is negotiated by representatives of the Bank and representatives of the potential loan recipient country, common provisions include privatization of the economy, maintaining a low rate of inflation and price 2 The Shining Path insurgency was another major factor increasing violations of physical integrity rights in Peru during the 1980s.

4 236 The Human Rights Effects of World Bank Structural Adjustment stability, shrinking the size of its state bureaucracy, maintaining as close to a balanced budget as possible, eliminating and lowering tariffs on imported goods, getting rid of quotas and domestic monopolies, increasing exports, privatizing state-owned industries and utilities, deregulating capital markets, making its currency convertible, and opening its industries and stock and bond markets to direct foreign ownership and investment (Meyer 1998). Good governance emphases of the Bank include eliminating government corruption, subsidies, and kickbacks as much as possible, and encouraging greater government protections of human rights including some worker rights (Sensor 2003; Kaufmann, Kraay, and Mastruzzi 2005). Most of the previous research has examined the IMF and its impacts, neglecting the role of the World Bank in promoting structural adjustment. Both are important actors, over the period examined in this study, the World Bank entered into 442 SAAs, while the IMF made The remainder of the article briefly reviews previous work on the economic effects of structural adjustment, elaborates on the theory briefly outlined above, discusses the earlier research estimating the impact of structural adjustment on human rights; elaborates upon the need for a selection model, presents some specific hypotheses, and provides evidence supporting those hypotheses. Finally, the theoretical, methodological, and policy implications of these results are discussed. The Economic Effects of Structural Adjustment The purpose of structural adjustment programs is to encourage economic growth (e.g., Harrigan and Mosley 1991; Przeworski and Vreeland 2000). According to neoliberal economic theory, structural adjustment programs reduce the size and role of government in the economy. A minimalist state produces and encourages economic growth, which promotes economic and social development (Chenery and Strout 1966). Limited government empowers individuals by giving them more personal freedom, making it more likely that all individuals will realize their potential. The ability to realize one s potential, according to this line of reasoning, leads to individual responsibility and self-reliance. Limited government maximizes individual opportunities, limits the opportunity for corruption and releases talented people into the more efficient private sector (Friedman 1962). Many scholars have examined the link between structural adjustment policies and economic growth and the weight of the evidence so far is that structural adjustment is not effective (Harrigan and Mosley 1991; Rapley 1996; Przeworski and Vreeland 2000; van de Walle 2001; Vreeland 2003). According to critics, the Fund and Bank use a conception of development that is too focused on economic growth, have misdiagnosed the obstacles to development in less developed countries, have failed to appreciate the value of government interventions into the private economy, and have insisted that structural adjustment reforms be implemented too quickly (Stiglitz 2002). It is possible that developing countries like China have been more successful, both in terms of aggregate economic growth and poverty reduction, because they have avoided SAAs from the IMF and World Bank. Unlike Russia, which has received a number of SAAs, China has avoided a rapid increase in economic inequality (Stiglitz 2002). Theory: The Human Rights Effects of Structural Adjustment Direct Effects Figure 1 depicts the main causal arguments of the conventional neoliberal and more critical views of the direct and indirect effects of structural adjustment on the human rights practices of governments. The direct effects may be theorized as 3 The IMF data comes from Vreeland (2003).

5 M. RODWAN ABOUHARB AND DAVID L. CINGRANELLI 237 a Rapid Economic Liberalization b Higher Level of Economic Development d c e More Respect for Physical Integrity Rights More Democracy f g More Respect for Economic Rights FIG. 1. Structural Adjustment and Human Rights: The Neoliberal Perspective positive or negative. The positive argument (linkage a ) is that a relatively limited government as required by SAAs is fundamental to all human freedoms. Limited government reduces barriers to the functioning of the free market, allowing people to enhance their opportunities and better pursue their own interests that are likely to be lost if human freedom is restricted (Friedman 1962; Hayek 1984). Consistent with this line of thought, Cranston (1964) has argued that respect for most human rights, including physical integrity rights (such as the right not to be tortured) only requires forbearance on the part of the state. However, as linkage h of Figure 2 indicates, structural adjustment programs also may have the direct effect of worsening government human rights practices, because a substantial involvement of government in the economy is essential for the protection of all human rights (Donnelly 2003). The historical record demonstrates, for example, that a reduced role of the state in capitalist economies has led to less protection of some human rights such as worker rights. From a principal-agent theoretical perspective, reducing the size of government also reduces the ability of principals (government leaders) to constrain the discretion of agents (police and soldiers). More administrative discretion is likely to lead to greater abuse of physical integrity rights (Policzer 2004). Also, in practice, the acceptance of structural adjustment conditions by the governments of less developed countries causes the adoption of new policies and practices. These new policies are designed to produce substantial behavioral changes in the affected populations. Evidence from literature about human learning suggests that people have a natural tendency to resist making substantial changes in their previous behavior (Davidson 2002). One of the tools government may use to overcome such resistance is coercion. The idea that liberalization and economic development may conflict with respect for some human rights is an enduring theme in the debate over development policy and an implicit element of structural adjustment packages. Loan recipient governments are expected to reduce their efforts to protect the social and economic h Rapid Economic Liberalization i Less Respect for Economic Rights j m More Domestic Conflict l n k Less Respect for Physical Integrity Rights Less Democracy FIG. 2. Structural Adjustment and Human Rights: The Critical Perspective

6 238 The Human Rights Effects of World Bank Structural Adjustment rights of their citizens in a variety of areas such as housing, health care, education, and jobs at least in the short run, with the expectation that they will be able to make much larger efforts toward these ends later. Civil and political liberties may have to be curtailed in order to ease the implementation of loan conditions (Donnelly 2003: ). People opposed to the policies of structural adjustment such as members of the press, trade unionists, leaders of opposition parties, clergy, social activists, and intellectuals may then be subjected to abuse of their physical integrity rights. Indirect Effects Figure 1 also depicts the expected indirect effects of structural adjustment on the human rights practices of loan recipient governments. As noted, neoliberal economic theory suggests that structural adjustment will promote economic development (linkage b in Figure 1). 4 Many previous studies (e.g., Poe, Tate, and Camp Keith 1999; Milner, Leblang, and Poe 2004) have shown that wealthier states have provided greater levels of respect for a wide variety of human rights including physical integrity rights (linkage c ). Thus, if the imposition of a SAA increases the level of wealth in a less developed country, then the indirect effect of SAA implementation should be an improvement in the human rights practices of governments. Despite findings showing that structural adjustment has not led to faster economic growth, the empirical debate over linkage b will continue. Thus, it is still important to understand the remainder of the neoliberal argument. As is indicated by linkages d and e in Figure 1, previous research has shown that wealthier states are more likely to be democratic (e.g., Lipset 1959; Przeworski et al. 2000; Boix 2003; Boix and Stokes 2003), and relatively high levels of democracy are associated with a higher level of respect for most human rights including physical integrity rights (Mitchell and McCormack 1988; Poe, Tate, and Camp Keith 1999; Davenport and Armstrong 2004; Milner, Leblang, and Poe 2004). Therefore, if the imposition of a SAA promotes higher levels of democratic development through increased wealth, then an indirect consequence of SAA implementation should be an improvement in human rights practices. Neoliberal defenders of the effects of SAAs on government respect for economic human rights have argued that higher levels of economic development caused by the implementation of a SAA will lead to improvements in government respect for economic rights (linkage g ) through what is now commonly referred to as the trickle down effect. That is, wealth will accumulate faster under a structural adjustment program, and, once accumulated, will trickle down to help the less fortunate in society. A number of studies have shown that the level of economic development has a strong, positive impact on basic human needs fulfillment (Moon and Dixon 1985; Rosh 1986; Spalding 1986; Park 1987; Milner, Poe, and Leblang 1999; Milner 2000; Milner, Leblang, and Poe 2004). Moreover, as indicated by linkage f, previous research has shown that democratic governments have been shown to make greater efforts to provide for the economic human rights of their citizens (Moon and Dixon 1985; Milner, Poe, and Leblang 1999; Milner, Leblang, and Poe 2004). Unfortunately, all of indirect neoliberal arguments linking SAAs to better human rights practices depend upon supporting evidence for linkage b in Figure 1. Without linkage b all of the other indirect causal chains from rapid economic liberalization to better human rights practices by governments are broken. At an earlier point in time, one might have argued that it was too soon to conclude that there was no evidence that the implementation of SAAs led to the accumulation of more wealth by loan recipients, but SAAs were initiated by the World Bank in 1980 and the IMF has had conditionality associated with its loans as far back as For a review of literature developing this argument, see Rapley (1996).

7 M. RODWAN ABOUHARB AND DAVID L. CINGRANELLI 239 (Sidell 1988). If SAAs have had a stimulative effect on economic development, it should be observable by now. The indirect effects posited by the critical perspective are summarized in Figure 2. There is a large body of research showing that implementation of a SAA has negative effects on government respect for economic human rights (linkage i ). Rapid economic liberalization, according to many observers, forces loan recipient states to reduce or even stop making efforts to help their citizens enjoy internationally recognized rights to health care, education, food, decent work and shelter, because structural adjustment conditions almost always require reductions on government spending for social programs (World Bank 1992; Chipeta 1993; Sowa 1993; Handa and King 1997; Meyer 1998; Zack-Williams 2000; Fields 2003). Some studies have emphasized the disproportionate negative economic human rights consequences for women (Commonwealth Secretariat 1989; Elson 1990; Buchmann 1996; Sadasivam 1997), for public sector employees and low-wage workers (Daddieh 1995). The poor and those in the public sector have seen their wages fall in real terms (Munck 1994; Daddieh 1995), while at the same time they have faced increased living costs because of the removal of price controls and subsidies for essential commodities (Zack-Williams 2000). The implementation of SAAs also has worsened the relative position of the worst off by increasing income inequality (Daddieh 1995; Handa and King 1997; Friedman 2000). Less attention has been given to the relationships explicitly linking the implementation of SAAs to subsequent government respect for physical integrity rights. 5 As shown in Figure 2, there are three indirect causal paths that should be considered (linkages jfk, j lfn, and m n ). All lead to less respect for physical integrity rights, and all depend upon empirical support for linkage i, which is plentiful. One line of thinking is that, by causing loan recipients to reduce their respect for the economic human rights of their most vulnerable citizens, externally imposed rapid economic liberalization of the type required by a SAA promotes domestic conflict (linkage j ), which, in turn, leads loan recipient governments to become more repressive (linkage k ). Acceptance of SAA conditions requires that decision makers in loan recipient countries enact unpopular policies. These policies cause hardships, especially among the poorest citizens, who are most dependent upon social programs (Vreeland 2002). Citizens, often led by organized labor, protest against reductions in social welfare programs and public employment, commonly required in SAAs (Pion- Berlin 1983, 1984). Sometimes the protests become violent (Auyero 2001; Fields 2003). The adjustment process also has intensified regional and ethnic conflicts as groups compete for a dwindling share of the national cake (Zack-Williams 2000:64). Increased repression (linkage k ) by the recipient government is one tool by which it can deal with violent protest (Davenport 1995; Fields 2003). However, it is important to distinguish incremental economic liberalization that results from a societal choice without undue external interference and pressure from the kind of rapid economic liberalization required by SAA conditionality. Economic liberalization that is not required by the conditions found within a SAA may not affect or may actually reduce domestic conflict in societies. For example, Hegre, Gissinger, and Gleditsch (2003) examine the impact of economic liberalization and find no discernable impact on the probability of civil conflict. Other critics of structural adjustment would like the Bank and Fund to give greater attention to the impacts of SAAs on issues such as democratic development (Pion-Berlin 1984; Stiglitz 2002). Increased domestic conflict caused by the implementation of SAAs presents serious challenges to democratic systems (linkage l ). Also, as indicated by linkage m, requiring democracies to enact unpopular pol- 5 There is a large body of literature from a dependency theory perspective arguing that rapid economic liberalization can worsen government human rights practices. For an excellent review of this literature, see Richards, Gelleny, and Sacko (2001).

8 240 The Human Rights Effects of World Bank Structural Adjustment icies, the Bank and Fund may be undermining democratic systems (Haggard 1995; Fields 2003). The positive relationship between a state s level of democracy and its respect for all types of human rights (linkage n ), as noted above, is well established in the literature. Thus, any policy that undermines democracy, undermines government respect for human rights. Previous Research Linking Structural Adjustment to Human Rights Practices The results of previous research explicitly focusing on the effects of SAAs on government respect for physical integrity rights are consistent with the expectations of the critical perspective (Franklin 1997; McLaren 1998; Camp Keith and Poe 2000; Fields 2003). Camp Keith and Poe (2000) evaluated the human rights effects of getting a SAA from the IMF by comparing the human rights practices of governments with and without such loans while controlling for other factors reliably associated with good or bad human rights practices by governments. They focused on a global sample of countries between 1981 and 1987, and found some evidence indicating an increase in the level of repression of physical integrity rights during the implementation of a SAA. Using a cross-sectional analysis, Franklin (1997) also found some support for the argument that governments implementing IMF agreements were likely to become more repressive. Furthermore, Camp Keith and Poe (2000) hypothesized that the very act of negotiating or entering into a loan with the IMF would have a temporary negative impact on the human rights practices of loan recipients. They were not clear about the rationale for this hypothesis, and their findings provided no statistically significant evidence for a negotiations effect. Others have argued that the involvement of international actors has a moderating effect on domestic conflicts (Grove 2001), which should have the effect of improving government respect for physical integrity rights. There also is a specific reason to expect that negotiating a SAA from the World Bank would have at least a temporary positive impact on the human rights practices of loan recipient governments. The U.S. International Financial Assistance Act in 1977 requires U.S. government representatives on the decision making boards of the World Bank and IMF to use their voices and votes to advance the cause of human rights in loan recipient countries (Abouharb and Cingranelli 2004a). The size of U.S. contributions to the Bank gives it a strong voice in loan negotiations (Banks, Muller, and Overstreet 2003). Thus, one would expect the World Bank to make SAAs with countries that have good human rights practices. Previous research has examined the effects of structural adjustment on the overall level of government respect for physical integrity rights but has not disaggregated the effects on torture, political imprisonment, extra judicial killing, and disappearances. 6 However, it is likely that the impacts of negotiating and implementing a structural adjustment program affect government respect for these kinds of physical integrity rights in different ways. In this early stage of the research program designed to develop theories explaining the human rights practices of governments, aggregate measures may mask theoretically important variations in how governments respect the human rights of their citizens (McCormick and Mitchell 1997). Disaggregating the measures of respect for physical integrity rights allows the investigation of whether governments improve or decrease their respect for different types of physical integrity rights to the same extent as a result of making and implementing a SAA from the World Bank. Existing theories explaining why governments resort to violent forms of political repression conceive of repression as the result of conscious choices by rational, utility maximizing political leaders (Poe and Tate 1994; Gartner and Regan 1996; 6 Camp Keith and Poe (2000), for example, used the five-point Political Terror Scale (PTS) to measure the degree of overall violation of those rights.

9 M. RODWAN ABOUHARB AND DAVID L. CINGRANELLI 241 Poe, Tate, and Camp Keith 1999). Both the domestic and international costs and benefits of violating different types of physical integrity rights vary. Torture and political imprisonment are the most common forms of physical integrity rights abuse by governments (Cingranelli and Richards 1999a). If government decision makers are rational, then policies allowing for the practice of torture and political imprisonment must offer higher net benefits than policies allowing the police or military to make citizens disappear or to kill them without a judicial process. If repression is a rational response to structural adjustment, then torture and political imprisonment should increase the most during the implementation of structural adjustment conditions. Since the end of the Cold War, however, there has been an increase in average worldwide government respect for the right against political imprisonment (Cingranelli and Richards 1999b). This trend indicates that, over time, either the costs associated with this form of repression have increased, the benefits have declined or both. The Need for a Selection Model Estimating the human rights effects of structural adjustment requires the use of a twostage econometric model. As explained by Achen (1986), Heckman (1988), Przeworski and Vreeland (2000), and Vreeland (2002, 2003) issues of endogoneity, selection, and randomization must be accounted for when assessing the impact of any public policy. One needs to disentangle the impacts of the policy from any prior attributes that may also have an impact (Collier 1991). In the context of the present problem, one must be able to distinguish whether the negative effects on physical integrity rights practices found by McLaren (1988) and Franklin (1997), Camp Keith and Poe (2000), were the result of the economic difficulties that made the loan recipient country a good candidate for a SAA in the first place or were they the consequence of the SAA itself. Single-stage models cannot provide an answer to that question. Single-stage models, like those used in previous cross-national studies of the impact of IMF conditionality on human rights practices also implicitly assume a unidirectional causal relationship. That is, structural adjustment loans affect human rights practices. More likely, human rights practices affect the probability of loan receipt, while loans affect human rights practices, which, in turn, affect the subsequent probability of loan receipt. Thus, both SAA receipt and human rights practices are mutually dependent or endogenous variables. Application of a singlestage model to estimate these theoretical relationships will generate inconsistent parameter estimates (Gujarati 1995). The methodological resolution to this conundrum is found in a variety of two-stage econometric models that disentangle the impact of these mutually dependent variables. 7 Which Countries Enter into SAAs? Through its public policy statements, the Bank has announced some of the criteria it uses to decide which governments should receive SAAs and which should not. The Bank s code of practice recommends that preference be given to applicants that are poor, have a capitalist ideology, have not nationalized private industry without providing fair compensation to the owners, are not able to borrow on the private market, and are creditworthy (Van de Laar 1980). These criteria created a bias against reaching agreements with communist countries, though some communist countries including the formerly communist Yugoslavia and Romania did receive them. In making its decisions, the Bank s Board of Directors must 7 Examples of selection models in research on human rights are rare. Blanton (2000) used a Heckman two-stage selection model to determine whether the promotion of human rights and democracy were important objectives affecting the decisions by the U.S. government to transfer arms abroad.

10 242 The Human Rights Effects of World Bank Structural Adjustment prioritize, sometimes among conflicting criteria, and operationally define terms such as capitalist ideology and creditworthy. Conventional wisdom holds that governments accept structural adjustment conditions because they face economic difficulties and need an infusion of foreign capital. 8 This means that they must sacrifice sovereignty over their economic policy (Bird 1996; Krasner 1999; Moyo 2001). There is significant disagreement over the roles of indicators of economic difficulty such as a large balance of payments deficit, while there is more agreement over the impact of other factors like lower foreign currency reserves, overvalued exchange rates, and negative changes in gross domestic product (GDP) which increase the probability of participation in structural adjustment programs (Vreeland 2003:12). Moreover, none of the purely economic explanations do a very good job of explaining participation in structural adjustment programs. Economic factors are part of the explanation of which governments receive SAAs, but they do not provide a complete picture. Non-Economic Selection Criteria Besides economic selection criteria, a variety of political, institutional, and social characteristics of potential recipient governments also affect the probability of reaching a SAA (Joyce 1992; Abouharb and Cingranelli 2004a, 2005). The Bank s Board of Directors decides which governments receive World Bank loans. The World Bank uses a weighted voting system for determining which agreements are approved and which are denied. The weights assigned are roughly in proportion to the share of the Bank s development funds contributed by each of the member governments. For the last 25 years, the United States and Japan have been the largest contributors to the Bank (Banks, Muller, and Overstreet 2003), so it is reasonable to assume that the preferences of their country representatives have dominated the preferences of other members of the Bank s Board of Directors. World Bank representatives protest against any allegations that their lending policies are motivated by political considerations, but the internal decision making process of the World Bank privileges the ideological perspectives of some governments over others, allows for logrolling and vote trading, and in all other respects provides fertile ground for what, in any other context, would be called politics. Despite this potential for politics, non-economic selection criteria have received relatively little attention. Some suggest that, unlike the IMF, the World Bank may prefer to work with governments willing to respect worker rights. Nelson (2000) contends that the Bank has in fact had a long-standing commitment to maintaining labor standards, because Bank officials believe that respect for three core labor standardsfagainst child labor, forced labor, and discrimination in hiring and treatment at workfactually promotes economic growth (Sensor 2003). In contrast, others suggest that structural adjustment conditions provided indirect incentives to limit worker rights in order to make countries more competitive internationally. The establishment of export processing zones are encouraged by the World Bank (Klak 1996:358). In an effort to make these zones as competitive as possible, the governments of developing countries attempt to keep wages low (Klak 1996:358). Thus, labor loses out in order to make countries as attractive as possible to international investors. Research investigating these competing claims on a large-n comparative basis has found evidence that the Bank is more likely to enter into agreements with countries that have higher levels of respect for worker rights (Abouharb and Cingranelli 2004a, 2005). Another non-economic factor alleged to increase the probability of participation in World Bank structural adjustment programs is an alliance with the United States 8 For other explanations of participation by the governments of developing countries in structural adjustment agreements, see Vreeland (2003).

11 M. RODWAN ABOUHARB AND DAVID L. CINGRANELLI 243 (Forsythe 1987). Recent work examining IMF selection criteria has argued that countries with policy preferences similar to key Fund contributors were more likely to receive preferential loan conditions (Stone 2004). By implication they also would be more likely to negotiate a SAA with the World Bank. Other work has found that being poor, having a large population (Abouharb and Cingranelli 2004a, 2005), and the end of the end of the Cold War (Williams 1994; Abouharb and Cingranelli 2005) increase the probability of entering into a SAA. The end of the Cold War marked the beginning of a period when international institutions, including international financial institutions, began to play a larger role in international affairs. Involvement in international war and high levels of domestic unrest have been found to reduce the probability of loan receipt (Abouharb and Cingranelli 2004a, 2005). A Disadvantage for Democracies? Several studies have found that more democratic governments were less likely to enter into SAAs (Pion-Berlin 1984; Przeworski and Vreeland 2000; Vreeland 2003). Putnam s (1988) theory of two-level games provides an explanation for the finding in the literature that the IMF prefers lending to authoritarian regimes. Putnam suggests that negotiations between an international agency like the World Bank or IMF and the leaders of a nation state can be thought of as a two-level negotiation game. Level I negotiation occurs between the leaders of the Bank and the leaders of the potential loan recipient country. Level II is played between the country leaders and their citizens. At level I, the leaders of the World Bank behave as autonomous, unitary actors in the model. At the risk of oversimplification, the preferences of the Bank are that decision makers in recipient countries agree to a set of economic reforms, these reforms be implemented faithfully, the economy of the recipient country improve, and the loans be paid back in a timely fashion (Williamson 1990). Domestic opposition makes it harder to reach any agreement. Domestic opposition might arise as a result of interest group efforts and opposition political parties, electoral cycles, and even institutional arrangements requiring legislative approval of international agreements. Putnam (1988) contends that the greater the autonomy of country leaders at level I from influence by their level II constituents, the greater the likelihood of achieving an international agreement. At level I, the leaders of authoritarian states can negotiate with greater authority and independence from domestic forces at level II. A bias against democracies in the selection processes of the World Bank is, thus, a predicted outcome of the model. Democratic leaders prefer not to lose the support of their constituents, and Bank leaders prefer not to give loans with conditions that may not be implemented by the loan recipient. There is a contrasting theoretical argument suggesting that democracies have an advantage when negotiating international agreements, because their governments can make more credible commitments (Leeds 1999; Martin 2000). According to this perspective, the properties of democratic accountability and institutionalized cooperation afford democracies the ability to send clear and credible signals concerning their ability and willingness to cooperate. Supporting this line of argument, Dollar and Svensson (2000) show that democratic governments are much more likely to fulfill the structural adjustment commitments they make to the World Bank. Hypotheses In order to test hypotheses about the human rights impacts of SAAs, one must first account for the effects of World Bank loan selection criteria. As noted, previous research suggests that economic, political, conflict, and human rights factors help

12 244 The Human Rights Effects of World Bank Structural Adjustment determine the probability of receiving a SAA and also impact subsequent human rights practices. Governments are more likely to enter into a SAA if they have: H1: Greater economic difficulty. H2: Greater respect for the human rights of their citizens. H3: An alliance with a major donor to the World Bank. H4: Larger populations. H5: Negotiated after the end of the Cold War. H6: More authoritarian domestic institutions. H7: Lower levels of domestic unrest. H8: Lower levels of interstate conflict. Previous research also supports the following second-stage hypotheses concerning the human rights impacts of SAAs: H9: The level of respect for physical integrity rights increases during the year a SAA is negotiated (the negotiation hypothesis). H10: The level of respect for physical integrity rights decreases during the years SAAs are implemented (the implementation hypothesis). H11: The practices of torture and political imprisonment will increase more after entering into and implementing structural adjustment conditions than the practices of extrajudicial killing and disappearance (the differential effects hypothesis). Other studies have demonstrated that wealthier countries, more democratic countries, and countries with a British colonial experience tend to have governments that provide more respect for the physical integrity rights of their citizens. Countries with military governments, relatively large populations, relatively large population increases, high levels of domestic conflict, and involvement in interstate war tend to have governments that provide less respect for the physical integrity rights of their citizens (Poe, Tate, and Camp Keith 1999; Poe 2002). These factors will be included as control variables in the analysis. Research Design This study uses a cross-national, annual time-series data set comprised of all nations of the world having a population of at least 5,000,000 in The data span the time period from 1981 to During this period, the World Bank awarded a total of 442 SAAs to countries in our sample, with a GDP per capita as high as $13,200. For this reason, the analysis includes all countries in the world, not just less developed countries. The unit of analysis is the country year. At the human rights impact stage we investigate whether entering into a SAA with the World Bank in a particular year or the implementation of those loan conditions in subsequent years have an impact on the probability of torture, political imprisonment, extra-judicial killing, and disappearances in loan recipient countries. Entering into a World Bank SAA is both a dependent variable in the first stage of the analysis and an independent variable in the second stage. It is a dichotomous measure that indicates whether a country received a World Bank SAA or not in a particular year. It is coded 1 for the years an agreement was made and 0 for all other years. The authors gathered the information necessary for constructing this measure from correspondence with officials at the World Bank. The measure of implementation of a World Bank SAA, an independent variable in the second stage, was generated by the authors. As most adjustment packages last for 3 years and the World Bank has determined that on average it takes 18 months for implementation to affect the economy, the results of the adjustment process should appear in years 2, 3, and 4 of the loan period (Jayarajah, Branson, and Sen 1996). For this reason, years 2, 3, and 4 after loan receipt were coded as 1 and otherwise as 0. It was assumed that entering into a SAA was followed by

13 M. RODWAN ABOUHARB AND DAVID L. CINGRANELLI 245 implementation of the negotiated structural adjustment conditionsfan assumption that is common in previous research examining the consequences of structural adjustment. However, using private World Bank records, Dollar and Svensson (2000) estimate that about one third of loan recipients do not fully implement the adjustment criteria demanded by the Bank so there is some variation in the effectiveness of implementation practices by loan recipient governments 9 that could not be captured by the measure of implementation used in this study. Even so, there is no reason to believe that the cases mistakenly coded as 1 for implementation rather than 0 generates systematic error in the empirical analysis. The measurement error generated creates a bias toward a weaker relationship than might actually exist between implementation and human rights practices of governments, but does not affect the direction of the relationship observed. Another limitation is that there was no information available about the particular structural adjustment conditions associated with each loan. The implementation of some provisions may have had greater human rights impacts than others, but, except for details contained in intensive case studies, the specific conditions imposed on loan recipients are not matters of public record. The human rights practices of governments are the dependent variables in the second stage. Four physical integrity rights from the Cingranelli and Richards ([CIRI] 2004) human rights data set were used as dependent variablesfextrajudicial killings, disappearances, political imprisonment, and torture. The sources of information used to develop this data set were the annual U.S. State Department Country Reports on Human Rights Practices and Amnesty International annual reports. Each of the four physical integrity variables was coded on a three-part scale where 0 ¼ frequent violations of the right (50 or more), 1 ¼ some violations (1 49), and 2 ¼ no violations. The correlations among the four physical integrity rights during the period ranged from a low of 0.27 between torture and disappearances to 0.49 between disappearances and extra-judicial killing. Among the independent variables pairwise correlations indicate no problems of multicollinearity. The highest pairwise correlations are found between worker rights and democracy at 0.62 and GDP per capita and democracy at The negotiation and implementation of SAAs are correlated at Tables 1 and 2 provide a summary of the operationalization of the independent variables used in the first and second stages of the analysis. The measure of overall respect for physical integrity rights used in the first-stage analysis is the CIRI physical integrity scale, a nine-point scale, ranging from zero, indicating no respect for physical integrity rights, to eight, indicating full respect for those rights. The worker rights variable used in the first-stage equation also was taken from the CIRI data set. It measures government respect for freedom of association at the workplace, the effective recognition of the right to collective bargaining, the elimination of all forms of forced or compulsory labor, the effective abolition of child labor, and acceptable conditions of work. This list is much the same as the International Labour Organization s list of five core labor rights. This variable also was coded on a three-point scale as follows. Worker rights are (0) not protected by the government, (1) somewhat protected by the government, (2) protected by the government (Cingranelli and Richards 2004). The theory of how structural programs affect the human rights practices of SAL recipients posits both direct and indirect effects. However, this research design only estimates the direct effects of negotiation and implementation of these programs while controlling for the effects of loan selection and other causal variables (e.g., level of democracy, level of economic development, and level of domestic conflict) in the theoretical model. This specification of the model is consistent with previous 9 Also, see Stone (2004).